Compare / ISO, NIST and the EU
COBIT 7 and AI governance
AI governance now rests on three instruments that COBIT 7 names but does not replace: a certifiable management system standard, a voluntary risk framework and a binding EU regulation. This page sets out what each one does, where COBIT 7 fits around them and where it leaves gaps you have to fill from the others.
The short answer
COBIT 7 is a general governance system for all enterprise information and technology, with no objective or practice dedicated to AI. ISO/IEC 42001 sets requirements for an AI management system, the NIST AI RMF gives voluntary AI risk outcomes, and the EU AI Act imposes legal obligations on AI providers and deployers.
In COBIT 7, AI appears inside activities, in new AI risk scenarios in the I&T risk profile design factor and in the IT implementation methods design factor. Appendix C newly lists the EU AI Act, ISO/IEC 42001 and the NIST AI RMF as referenced standards. That is recognition, not a mapping.
The practical split: COBIT 7 provides the board accountability, roles, risk appetite, compliance and assurance structure around AI. The other three provide the AI-specific content: impact assessment, data and bias controls, human oversight, transparency and, for the AI Act, conformity assessment and fines.
At a glance
| COBIT 7 | ISO/IEC 42001:2023 | NIST AI RMF 1.0 | EU AI Act | |
|---|---|---|---|---|
| Owner | ISACA | ISO and IEC (JTC 1/SC 42) | NIST, US Department of Commerce | European Parliament and Council |
| Current version | COBIT 7, released 19 November 2026 | Edition 1, December 2023, no amendment | Version 1.0 (NIST AI 100-1), 26 January 2023, revision announced | Regulation (EU) 2024/1689, amended by Regulation (EU) 2026/1744 |
| Type of document | Governance and management framework for enterprise I&T | Management system standard with auditable requirements | Voluntary risk management framework | Binding regulation, directly applicable in all Member States |
| AI-specific | No. AI sits inside activities and design factors | Yes | Yes | Yes |
| Scope | All enterprise information and technology | The AI management system of an organization that develops, provides or uses AI | AI risks and trustworthiness across the AI life cycle | AI systems and general-purpose AI models placed on the market, put into service or used in the EU |
| Structure | 5 domains, 40 objectives, 231 practices, 2,507 activities, 13 design factors | Clauses 4 to 10, Annex A with 38 controls in 9 areas, Annexes B to D | 4 functions, 19 categories, 72 subcategories, 7 trustworthiness characteristics | Risk tiers, obligations by operator role, Annexes I and III |
| Who must act | Board and management, by choice | The organization that adopts it | Any organization that chooses to use it | Providers, deployers, importers, distributors and others, including non-EU operators |
| Certification of organizations | None. Certificates are for people (COBIT 7 Foundation) | Voluntary, by certification bodies assessed under ISO/IEC 42006 | No NIST certification scheme | Conformity assessment and CE marking for high-risk systems |
| How performance or conformity is shown | Capability levels 1 to 5 and three assessment tiers | Third-party audit against the requirements | Profiles and self-attestation | Conformity assessment, registration, market surveillance; harmonized standards give presumption |
| Consequence of failure | None external | Loss of certificate | None external | Fines up to EUR 35 million or 7% of worldwide annual turnover |
| Cost and access | Free basic digital access, premium content paid | Paid, CHF 225 on iso.org | Free | Free on EUR-Lex |
How they differ
General governance versus AI-specific requirements
COBIT 7 governs all I&T and treats AI as one technology among others. Its 40 objectives keep the COBIT 2019 names and none of them is about AI. ISO/IEC 42001, the AI RMF and the AI Act are written for AI and contain requirements, outcomes or obligations that COBIT has no counterpart for, such as AI system impact assessment and human oversight.
Voluntary frameworks versus binding law
COBIT 7, ISO/IEC 42001 and the NIST AI RMF are voluntary. The EU AI Act is law: it applies to providers placing AI on the EU market wherever they are established, and to non-EU providers and deployers whose AI output is used in the EU. Fines reach EUR 35 million or 7% of worldwide annual turnover for prohibited practices.
Certification and assurance
COBIT 7 certifies people only and assesses organizations through capability levels and three assessment tiers. ISO/IEC 42001 can be certified by independent certification bodies, which are themselves assessed under ISO/IEC 42006:2025. NIST runs no certification for the AI RMF. The AI Act uses conformity assessment and CE marking for high-risk systems.
Unit of analysis
COBIT 7 looks at the enterprise governance system as a whole. ISO/IEC 42001 looks at the organization's AI management system. The AI RMF works at the level of the organization and of each AI system. The AI Act attaches obligations to each AI system or model and to each operator role, such as provider or deployer.
What COBIT 7 adds for AI
ISACA lists three AI changes in COBIT 7: new AI risk scenarios in the I&T risk profile design factor, IT implementation methods that consider AI and cloud, and refreshed IT-related issue examples. Inside the objectives, one activity in BAI02.03 covers AI-specific requirements risk, and other activities use AI as a tool for monitoring, security, data and assurance.
How current each one is
COBIT 7 was released on 19 November 2026. ISO/IEC 42001 is still its 2023 first edition, adopted in Europe as EN ISO/IEC 42001:2026 with identical text. NIST says the AI RMF 1.0 is being revised but has published no new version. The AI Act was amended in July 2026, which moved the high-risk dates.
Where they overlap
No mapping between COBIT and ISO/IEC 42001, the NIST AI RMF or the EU AI Act has been published by ISACA, ISO, NIST or the EU, so every row below is our own reading of COBIT 7 objective names against each instrument. Where ISACA's January 2025 white paper on COBIT for AI links the same theme to an objective, the note says so; that paper maps trustworthy AI themes to COBIT 2019 objectives, not clauses, functions or articles. Treat the table as a starting point for your own mapping.
| Instrument | Part | COBIT 7 objectives | Origin |
|---|---|---|---|
| ISO/IEC 42001 | Clause 5: leadership, AI policy, roles and responsibilities | Our reading | |
| ISO/IEC 42001 | Clauses 6.1, 8.2 and 8.3: AI risk assessment and treatment ISACA's white paper links AI risk as a theme to EDM03 and APO12. | Our reading | |
| ISO/IEC 42001 | Clauses 6.1.4 and 8.4: AI system impact assessment No COBIT equivalent. The nearest is the BAI02.03 activity on AI-specific requirements risk. | Our reading | |
| ISO/IEC 42001 | Clause 7: resources, competence, awareness, documented information | Our reading | |
| ISO/IEC 42001 | Clause 9: monitoring, internal audit, management review ISACA's white paper links monitoring as a theme to MEA01 and MEA02. | Our reading | |
| ISO/IEC 42001 | Annex A.6: AI system life cycle | Our reading | |
| ISO/IEC 42001 | Annex A.7: data for AI systems Covers data management and quality. Examination of data for bias has no COBIT counterpart. | Our reading | |
| ISO/IEC 42001 | Annex A.10: third-party and customer relationships | Our reading | |
| NIST AI RMF | Govern: policies, accountability, culture, workforce, third parties | Our reading | |
| NIST AI RMF | Map: context, categorization, impacts, benefits and costs Impact assessment in Map has no direct COBIT counterpart. | Our reading | |
| NIST AI RMF | Measure: testing, evaluation, verification, validation, trustworthiness metrics Fairness and bias measurement has no COBIT counterpart. | Our reading | |
| NIST AI RMF | Manage: prioritize, respond, monitor, handle incidents, improve | Our reading | |
| EU AI Act | Knowing the Act applies, inventory and tracking of obligations The legal classification of systems (prohibited, high risk, transparency risk) has no COBIT counterpart. | Our reading | |
| EU AI Act | Article 4: AI literacy of staff | Our reading | |
| EU AI Act | Article 9: risk management system for high-risk AI | Our reading | |
| EU AI Act | Article 10: data and data governance Data quality and life cycle only. Representativeness and bias examination are not covered. | Our reading | |
| EU AI Act | Articles 11 and 12: technical documentation and record-keeping | Our reading | |
| EU AI Act | Article 17: quality management system of providers Presumption of conformity for Article 17 comes from the harmonized standard EN 18286:2026, not from COBIT. | Our reading | |
| EU AI Act | Articles 72 and 73: post-market monitoring and serious incident reporting | Our reading |
Published mapping: taken from a mapping published by the framework owner or by ISACA. Third-party source: taken from a named publication. Our reading: our own comparison at objective level, not an official mapping. Select an objective to open its practices.
Using them together
A workable division of labor: the AI Act defines what is legally required of whom, ISO/IEC 42001 gives a certifiable management system for running AI responsibly, the NIST AI RMF gives a free vocabulary and list of outcomes for AI risk work, and COBIT 7 holds the enterprise governance system that the other three sit in. This is our reading; no source prescribes it.
COBIT 7 contributes what the AI instruments assume but do not build: board accountability for governance, the goals cascade, defined roles with one accountable role per practice, risk appetite through EDM03 and APO12, compliance with external requirements through MEA03 and assurance through MEA04. Use the AI risk scenarios in the I&T risk profile design factor when tailoring the governance system.
If you already run COBIT, add what it lacks: an AI system inventory with classification against the AI Act, AI system impact assessment, controls for data bias, human oversight, transparency and the model life cycle taken from ISO/IEC 42001 Annex A or the AI RMF, and the legal conformity route for any high-risk system.
Which one do you need?
COBIT 7 is enough to start when
- You need board-level accountability and roles for AI decisions
- AI is one technology among many in your I&T portfolio
- You want AI risk inside the existing enterprise risk process
- You need a compliance and assurance structure for AI obligations
You need an AI instrument when
- You place or use AI systems in the EU and the AI Act applies
- Customers or regulators ask for a certified AI management system
- You need impact assessment, bias testing or human oversight controls
- You want a free, outcome-based AI risk vocabulary (NIST AI RMF)
You need both when
- AI obligations must be owned at board level and evidenced in audits
- You run several frameworks and need one governance system over them
- You prepare high-risk AI systems for the December 2027 or August 2028 dates
Questions people ask
Does COBIT 7 have an objective for AI?
- No. COBIT 7 keeps the 40 objectives of COBIT 2019 with the same identifiers and names, and none is dedicated to AI. No AI focus area was published with COBIT 7. AI appears inside activities, mostly at capability level 5 as an enabling technology, and in one activity of BAI02.03 that covers AI-specific requirements risk such as model bias and adversarial manipulation. The broader AI content sits in the design factors: new AI risk scenarios in the I&T risk profile and AI in IT implementation methods.
Is there an official mapping from COBIT to ISO/IEC 42001, the NIST AI RMF or the EU AI Act?
- None was found on 22 November 2026. ISACA's white paper "Leveraging COBIT for Effective AI System Governance" (31 January 2025) maps trustworthy AI themes such as risk, security, privacy and monitoring to COBIT 2019 objectives, but not to clauses, subcategories or articles of the three instruments. COBIT 7 lists all three as referenced standards in Appendix C. NIST's list of AI RMF crosswalks includes no COBIT crosswalk.
What is the difference between ISO/IEC 42001 and the NIST AI RMF?
- ISO/IEC 42001:2023 is a management system standard whose requirements can be audited, with voluntary certification by independent certification bodies. It has clauses 4 to 10 and 38 reference controls in Annex A. The NIST AI RMF 1.0 is a free, voluntary set of outcomes organized in four functions (Govern, Map, Measure, Manage), 19 categories and 72 subcategories, with no certificate. They are compatible: NIST hosts a crosswalk between the two, provided by Microsoft.
Is ISO/IEC 42001 required by the EU AI Act?
- No. The AI Act does not require any standard, and the European Commission states that applying standards remains voluntary and providers may choose other ways to show compliance. Presumption of conformity comes only from harmonized standards cited in the Official Journal. On 22 November 2026 the only one is EN 18286:2026, a quality management system standard for AI Act purposes that covers Article 17. The Commission describes it as designed to be layered on top of ISO/IEC 42001 or another QMS, not to replace it.
When do the AI Act high-risk obligations apply?
- After the amendment by Regulation (EU) 2026/1744, the Digital Omnibus on AI, which entered into force on 27 July 2026, the high-risk rules apply from 2 December 2027 for systems listed in Annex III and from 2 August 2028 for AI in products covered by Annex I. Pages written before July 2026 often give older dates. Prohibitions, AI literacy and general-purpose AI rules were not postponed, and high-risk systems intended for public authorities must comply by 2 August 2030.
Does the EU AI Act apply to organizations outside the EU?
- Yes. Article 2 covers providers that place AI systems or general-purpose AI models on the EU market, wherever they are established, and providers and deployers in third countries where the output of the AI system is used in the EU. Importers, distributors, product manufacturers and authorized representatives of non-EU providers are also in scope. Fines reach EUR 35 million or 7% of worldwide annual turnover for prohibited practices, and EUR 15 million or 3% for most other obligations.
Is the NIST AI RMF being revised?
- Yes, but no new version exists. NIST states that the AI RMF 1.0 is being revised as part of the White House AI Action Plan and publishes no draft, version number or date, so version 1.0 (NIST AI 100-1) remains the one to cite. Following Executive Order 14434 of 29 September 2026, some NIST pages now use the term "super intelligence", while the framework page and publication identifiers are unchanged. The generative AI profile, NIST AI 600-1 of July 2024, remains published.
Can an organization be certified against COBIT 7 for AI governance?
- No. COBIT 7 has no certification of organizations. It offers three assessment tiers: board-level maturity, high-level achievement and detailed capability at levels 1 to 5, which can be applied to the objectives chosen for AI. Certificates are for people, starting with COBIT 7 Foundation. An organization that wants a certificate for how it manages AI would look at ISO/IEC 42001, whose certification is voluntary and carried out by independent certification bodies.
Which ISACA credentials cover AI?
- ISACA has launched three advanced AI credentials for individuals: AAIA, Advanced in AI Audit (May 2025, requires an active CISA or a qualified designation), AAISM, Advanced in AI Security Management (August 2025, requires an active CISM or CISSP), and AAIR, Advanced in AI Risk (April 2026). A fourth, AAIG, Advanced in AI Governance, is in beta with no launch date published. These credentials certify people. They are not COBIT products and say nothing about an organization's conformity.
Sources
- ISACA launches updated COBIT 7 · ISACA
- Leveraging COBIT for effective AI system governance · ISACA
- ISO/IEC 42001:2023 catalogue entry · ISO
- ISO/IEC 42001 explained · ISO
- AI Risk Management Framework · NIST
- AI RMF playbook · NIST AI Resource Center
- AI RMF crosswalks · NIST AI Resource Center
- Regulation (EU) 2024/1689 (AI Act) · EUR-Lex
- Regulation (EU) 2026/1744 (Digital Omnibus on AI) · EUR-Lex
- Timeline for the implementation of the AI Act · European Commission, AI Act Service Desk
- First harmonised standard supporting the AI Act published · European Commission
- Standardisation of the AI Act · European Commission
Checked on 22 November 2026. ISO/IEC 42001 is published by ISO and IEC, the AI RMF by NIST, and the EU AI Act is Regulation (EU) 2024/1689 of the European Parliament and of the Council. COBIT is a registered trademark of ISACA. Descriptions are our own wording. For anything you rely on, read the source.